Content marketing measurement: what changed since 2012

Around 2012, if you sat in a marketing meeting and asked how the blog was doing, you got a shrug and a page-view count. Content marketing was new enough that the serious question about content marketing measurement was whether it was possible at all, or whether content was a matter of faith, like sponsoring the local theater.

Sit in the same meeting now and the problem has flipped. There is a dashboard. It has forty tiles: sessions, scroll depth, time on page, shares, backlinks, assisted conversions. Nobody asks whether content can be measured. The question is why, with all this, nobody can say whether a customer was helped.

Content marketing measurement is the work of deciding, with evidence, whether a piece of content did what it was published to do, for the reader and for the business. That is what changed: we solved the wrong half of it. Most content metrics measure the content. Almost none of them measure the customer.

Key takeaways

• Content marketing measurement went from having no metrics to having too many, and most of the new ones describe the asset rather than the person who read it.
• Every content metric belongs to one of two piles, what happened to the content or what happened to the customer, and only the second pile says whether the content worked.
• A piece of content is an experience, so the questions a CX team asks about a checkout flow (did it reduce effort, did it resolve the need, did it shorten the path) apply to it directly.
• Two free-text survey questions, “how did you first hear about us” and “what almost stopped you”, measure content better than any attribution model.
• Support deflection is content measurement that already exists in most companies; it just lives in a different meeting.
• A five-line monthly scorecard that CX and marketing both sign keeps the customer-side measures from being crowded out by the tiles.

What is content marketing measurement

Content marketing measurement is deciding, with evidence, whether a piece of content did its job. The job is always the same at bottom: somebody arrived with a question, a doubt or a task, and the content either helped or it did not. Measurement is how you find out which.

That is not the same thing as a dashboard. A dashboard is a collection of numbers the analytics stack can produce without asking anyone. Measurement starts from the question “did it work” and goes looking for the evidence wherever it lives, including in places the analytics stack cannot see: the ticket queue, the sales call, the customer’s memory.

Sort the forty tiles into two piles.

In the first pile goes everything that describes what happened to the asset: it was viewed, scrolled, shared, ranked, linked. These are easy to collect, they move every day, and they can all go up while the customer gets nothing.

In the second pile goes everything that describes what happened to the person: did they find the answer, did they still need to call, did they buy with more confidence. These are harder to collect, they move slowly, and they are the only ones that matter.

Most content dashboards are entirely the first pile, because the analytics tools ship with the first pile and the second pile requires you to ask the customer. The familiar buckets those tiles fall into, and the bucket they miss, are laid out in content marketing metrics beyond the sale.

How to measure content like an experience

A piece of content is an experience. Somebody arrived with a question, a doubt, or a task, and the page either helped or it did not. So the usual customer experience questions apply, and each one can be turned into a measure without buying anything.

1. Ask whether it answered the question. Put a yes or no at the bottom of every answer page, “did this answer your question?”, with an optional comment. A short page that answers fully beats a long one that performs well. Read every comment on the no answers; they are the next content brief.
2. Measure effort. The customer wanted an answer. Did they get it on one page, or did they read three, open a chat, and wait? A help center visit followed by a ticket within the hour is the content equivalent of a high customer effort score.
3. Check whether it shortened the path. A buyer who reads a good pricing page arrives at the sales call with fewer questions. A user who reads a good help article does not open a ticket. Ask sales once a month which topics buyers now arrive knowing, and ask support which topics have gone quiet.
4. Ask “how did you first hear about us” as free text. Not a dropdown. The dropdown hands you your own categories back. The free-text version gives you the article title someone half-remembers, the podcast episode, the colleague who forwarded a guide. A month of answers tells you which content actually starts relationships.
5. Ask new customers “what almost stopped you”. Ask it after purchase, while the doubt is still fresh. Some answers will be about price, some about a competitor, and some about a question they could not find answered on your site. Each of those is a brief for content that answers real questions.
6. Put the results in front of both teams. The CX lead already asks these questions about the checkout flow. Content should not be exempt, and it should be reviewed in the same meeting.

A worked example (illustrative)

Take a help article on a topic that generated about 300 tickets a month. It went live on the first of the month. In the following month the topic generated 200 tickets, the article was viewed 2,000 times, and 70 percent of the people who answered the on-page question said it answered theirs.

The first pile says: 2,000 views. The second pile says: roughly 100 conversations that did not have to happen, and 30 percent of readers still stuck, with comments explaining why. The second pile is the one that tells you what to fix in the article and what to tell the support manager. The 2,000 views tell you nothing either of them can act on.

Observed, customer-reported and operational measures: which to use when

Content marketing measurement draws on three kinds of evidence, and each is blind to something the others see.

Kind of measure	Examples	What it tells you	What it misses	Use it for
Observed (analytics)	Sessions, scroll depth, time on page, click paths	What happened to the asset, in real time	Whether the reader got what they came for	Spotting what is being found and what is not
Customer-reported	“Did this answer your question?”, “how did you hear about us”, “what almost stopped you”	What the reader thinks happened, in their words	Precision; small samples; memory gaps	Deciding whether the content worked and what to write next
Operational	Ticket volume by topic, call length, agent link usage, sales notes	What changed in the business because of the content	Which piece caused it, unless you compare carefully	Proving the result to people outside marketing

Use all three, but let the bottom two rows decide. Operational and customer-reported measures say whether content worked; observed measures explain how people got there. When the rows disagree, believe the customer.

Support deflection is content measurement

The most measurable content most companies own is the help center, and it usually sits outside marketing’s dashboard.

When a help article is good, the ticket does not arrive. That absence can be counted: tickets on a topic before and after the article went live, the share of help center visits followed by a ticket within the hour, and how often an agent pastes a link instead of writing a reply. A drop in tickets is a content result visible in the cost line.

The before-and-after comparison has a known weakness: other things change in the same month. A product release, a billing cycle or a seasonal peak can move ticket volume on its own. The cleaner test, where volume allows, is to compare customers who saw the article with similar customers who did not, which is the same logic as measuring customer experience with control groups. Where the volume is too small for that, note the other changes alongside the drop and let the reader judge.

CX teams already track these numbers. Nobody calls it content measurement, so it never reaches the content meeting.

What quietly breaks content marketing measurement

Three habits undo good measurement, and each one looks reasonable from inside the team that has it.

Attribution that flatters the last click. Attribution tools give credit to the touchpoints they can see, and they see clicks. The last click before a purchase is the easiest to observe, so it gets the credit, even when the customer decided weeks earlier, after a comparison page a colleague sent them. The article that did the work is invisible; the branded search that came afterwards looks like a hero. Multi-touch models redistribute the credit, but only among observed clicks. The forwarded email, the hallway conversation, the article read on a phone and acted on from a laptop: none of it appears. The free-text “how did you hear about us” question is the only attribution model that includes what the customer remembers.

Measuring every piece the same way. A pricing explanation, a help article and a trend piece have different jobs. Judging all three by sessions makes the trend piece the winner every month and slowly moves the calendar toward trend pieces. Each piece needs the one measure that matches its job, decided before it is published.

Reporting to the wrong room. Content results that describe the customer belong in the meeting where customer results are reviewed. When they stay in the marketing dashboard, the ticket drop is credited to support, the shorter calls to sales, and the content team is left defending page views.

The older argument in is your program worth the effort? The role of measurement applies unchanged: measurement that only counts what is convenient steers the work toward what is convenient.

When content marketing measurement is not the answer

Some content should not be measured piece by piece, and pretending otherwise produces bad numbers and worse decisions.

Low volumes defeat most of the methods. A page read by forty people a month will not produce a usable helpfulness rate or a visible ticket change. Group such pages by topic and measure the topic, or accept that the evidence will be qualitative: what the six people who commented said.

Brand content has a job that is real and slow. A well-written piece that makes the company’s point of view clear will not shorten a sales call this quarter. Measuring it as if it should will kill it, and some of it deserves to live. Give it a different measure (unprompted mentions, “how did you hear about us” answers over a year) and a smaller share of the budget.

Measurement also lags. A buying decision that a comparison page helped along in March may close in September. Monthly scorecards show trends; they do not show the full return of any single month’s work, and the way buyers evaluate a purchase before they ever call makes the lag longer, not shorter.

And no measure replaces reading the comments. A helpfulness rate of 70 percent is a number. The thirty comments explaining the other 30 percent are the work.

A five-line scorecard two teams can sign

Here is a monthly scorecard short enough to fit in an email and honest enough for both the CX lead and the marketing lead to sign.

Line	What it counts	Owner
Questions answered	Customer questions (tickets, calls, site search) that now have a published answer	Marketing
Helpfulness	Share of yes answers to “did this answer your question?”, with the no comments read	CX
Tickets avoided	Change in ticket volume on topics with new or updated content	Support
Path shortened	Sales’ read of whether calls got shorter or better on covered topics	Sales
Customer-reported source	Free-text “how did you hear about us” answers that name a piece of content	Marketing and CX

Five lines, no sessions, no scroll depth. Each describes something that happened to a customer and can be checked against a source other than the analytics tool. The fourth line depends on sales taking part, and sales takes part when the program hands something back, such as the answer pages their prospects keep asking for.

The other forty tiles can stay on the dashboard for anyone who finds them useful. They just do not get anyone’s signature.

Where to start

1. Add the helpfulness question to your ten most-visited answer pages. Yes or no, optional comment. Read the comments weekly.
2. Change “how did you hear about us” to free text on the sign-up or post-purchase survey, and tag a month of answers by hand.
3. Ask support for ticket counts by topic for the last three months, and mark which topics have a published answer. That is the “tickets avoided” baseline.
4. Ask sales one question at the monthly meeting: which topics do prospects now arrive already knowing, and which are still eating call time?
5. Fill in the five-line scorecard for last month, gaps included. The gaps tell you which team you have not yet asked.
6. Retire ten tiles from the dashboard that nobody has acted on in a quarter. Nobody will notice, which is the point.

FAQ

What is content marketing measurement?

Content marketing measurement is deciding, with evidence, whether a piece of content did the job it was published for. The evidence comes from three places: analytics that observe the reader, surveys that ask the reader, and operational data such as ticket volume and sales call notes that show what changed in the business.

What metrics matter most for content marketing?

The metrics that describe what happened to the customer: whether the page answered their question, whether they still had to call, whether they arrived at the sales conversation better informed. Views, shares and scroll depth describe what happened to the asset and can all rise while the reader gets nothing.

How do you measure the ROI of content?

Count the conversations the content replaced (tickets avoided, shorter sales calls) and the customers who named it as their first contact with the company, then put those against the cost of producing it. Attribution reports based on clicks understate content’s share, because they miss forwarded links and remembered reading.

How do you know if content is helping customers?

Ask them, on the page, with a yes or no question and an optional comment. Then check the operational data: did tickets on that topic fall, and does sales say buyers now arrive knowing the answer? The two sources together are more reliable than either alone.

Why does last-click attribution undervalue content?

Last-click attribution credits the final observable click before a purchase, which is usually a branded search or a direct visit. The article that persuaded the buyer weeks earlier, often read on another device or forwarded by a colleague, leaves no click in the path and gets no credit.

How often should you review content metrics?

Monthly is enough for a scorecard that describes customer outcomes, because ticket volumes and sales feedback move slowly. Daily dashboards mostly show the observed metrics, which tempt teams to react to noise. Read the on-page comments weekly, since each one is a specific fix or a new brief.